{
  "fetched_at": "2026-10-19T23:28:39Z",
  "source_url": "https://www.kaggle.com/competitions/aisoc-regression-workshop/data",
  "pages": [
    {
      "name": "data-description",
      "content": "The dataset contains job postings from Glassdoor.com from 2017-2018, your task is to predict avg_salary based on the other features.\n\n## Files\n\n*   **train.csv** - the training set\n*   **test.csv** - the test set\n*   **sample_submission.csv** - a sample submission file in the correct format\n\n## Columns\n\n*   `id` - The unique identifier for the job posting\n*   `avg_salary` - Average salary for this listing (target)\n*   `job_state` - The state where the job is located\n*   `same_state` - A binary indicator of whether the job is in the same state as the person looking at the job\n*   `age` - The age of the person looking at the job\n*   `python_yn` - A binary indicator of whether the person looking at the job knows Python\n*  `R_yn` - A binary indicator of whether the person looking at the job knows R\n*   `spark` - A binary indicator of whether the person looking at the job knows Spark\n*   `aws` - A binary indicator of whether the person looking at the job knows AWS\n*   `excel` - A binary indicator of whether the person looking at the job knows Excel",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "Welcome to the \"Predicting Average Salary from Glassdoor Job Postings\" competition! In this data science challenge, participants are tasked with developing a predictive model that can estimate the average salary of job listings based on various features extracted from job postings on Glassdoor.com. The dataset provided contains a wealth of information from job postings spanning the years 2017 to 2018.\n\n## Key Features\n\n*   `id` - The unique identifier for the job posting\n*   `avg_salary` - Average salary for this listing (target)\n*   `job_state` - The state where the job is located\n*   `same_state` - A binary indicator of whether the job is in the same state as the person looking at the job\n*   `age` - The age of the person looking at the job\n*   `python_yn` - A binary indicator of whether the person looking at the job knows Python\n*  `R_yn` - A binary indicator of whether the person looking at the job knows R\n*   `spark` - A binary indicator of whether the person looking at the job knows Spark\n*   `aws` - A binary indicator of whether the person looking at the job knows AWS\n*   `excel` - A binary indicator of whether the person looking at the job knows Excel\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Objective",
      "content": "The objective of this competition is to develop predictive models that accurately estimate the average salary for job listings based on a diverse set of features extracted from Glassdoor job postings spanning the years 2017-2018. Participants are challenged to leverage their data science and machine learning skills to create models that provide valuable insights into salary expectations for job seekers and assist employers in offering competitive compensation packages. Competitors will be evaluated based on their ability to minimize the Mean Absolute Error (MAE) when predicting salaries. Join this competition and demonstrate your data science prowess in the realm of job market analytics!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": " - Don't cheat!\n - Apply yourself!\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "abstract",
      "content": "Welcome to the University of Southampton Artificial Intelligence Society's Linear Regression Workshop Kaggle Challenge! In this competition, participants are tasked with predicting the salary of an employee based on a set of diverse features. This is your chance to apply practical machine learning and showcase your regression prowess.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "**Evaluation Metric - Mean Absolute Error (MAE):**\n\nIn this competition, we will use Mean Absolute Error (MAE) as the primary evaluation metric to assess the performance of participants' predictive models. MAE is a widely recognized metric that quantifies the average absolute difference between the predicted average salaries and the actual salaries in the test dataset. The lower the MAE, the better the model's predictive accuracy.\n\n**How MAE is Calculated:**\n\nFor each prediction, the MAE is computed as the absolute difference between the predicted salary (y_pred) and the true salary (y_true). The MAE for all predictions is then averaged to obtain the final score:\n\nMAE = (1/n) * Σ |y_true - y_pred|\n\nWhere:\n- n is the number of predictions in the test dataset.\n- Σ denotes the summation across all predictions.\n\n**Interpreting MAE:**\n\nA lower MAE signifies that the model's predictions are closer to the true average salaries, indicating higher accuracy. Conversely, a higher MAE suggests that the model's predictions have larger deviations from the actual salaries, signifying lower accuracy.\n\n**Scoring:**\n\nYour model's MAE score will be used to rank your performance on the Kaggle leaderboard. The participant with the lowest MAE will achieve the highest rank, signifying the best predictive performance.\n\n**Key Takeaway:**\n\nTo excel in this competition, focus on developing models that minimize the Mean Absolute Error. This metric quantifies the average absolute difference between predicted and actual salaries, providing a reliable measure of predictive accuracy.\n\nGood luck, and may the most accurate predictive models rise to the top of the leaderboard!\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}